call
Go to Console

Guard Rail Console Guide

This page provides a guide to the Guard Rail entity in the Namirasoft Inference Console. It defines the concepts and configuration fields used when creating and managing Guard Rails. Use this guide to understand the purpose and behavior of each setting available during Guard Rail configuration.

What Is a Guard Rail?

A Guard Rail protects your workflow by screening content against the rules you enable. It can screen the user prompt, the AI output, or both, and a request stops with a clear error when a banning rule matches. Rules range from built-in safety and privacy categories, such as hate speech, violence, and personal data, to your own Conditions and prompts.

The Challenge a Guard Rail Solves

Everything a user types reaches the model, and everything the model says can reach the user, unless rules stand between them.

Common challenges include:

  • Unsafe content: Hate speech, violence, and other unsafe categories need detection on both sides of the exchange. Building that detection is specialized work.
  • Sensitive data exposure: Emails, phone numbers, cards, and credentials can leak into prompts or answers. Screening for them by hand is incomplete on the first day and outdated by the second.
  • Prompt injection: Crafted prompts try to override your instructions. Recognizing them is a moving target no single application should chase alone.

How Namirasoft Inference Solves the Problem

You add rules to the Guard Rail, and each rule has a type, an optional value, and two switches that decide where it applies, On User Prompt and On AI Output. Attach the Guard Rail to a Gateway, and every request that passes through is screened by all of its rules.

Overview of Guard Rail Fields and Options

Below is a detailed explanation of the fields available when creating or managing a Guard Rail. Understanding these fields helps ensure your Guard Rail is configured correctly for your requirements.

  • ID (String): This is a unique identifier automatically assigned to the Guard Rail when it is created. The system uses it to track, reference, and manage this specific Guard Rail. This value is auto-generated and cannot be modified.
  • User ID (Namirasoft Account’s ID): This is the unique identifier of the Namirasoft Account user who owns this Guard Rail. It is used internally for permission control, audit logging, and access management.
  • Workspace ID (Namirasoft Workspace’s ID): This is the identifier of the workspace this Guard Rail belongs to, as defined in Namirasoft Workspace. A workspace is a shared organizational space where teams group their work, projects, and members.
  • Name (String): This is a label used to identify this Guard Rail in the console. A good name states what it protects against. For example: “Safety and PII Guardrail”.
  • Rules (List): These are the protections this Guard Rail applies. Each rule has a Type, the value that type needs, and the On User Prompt and On AI Output switches that decide where it applies. The available types are:
    • Condition: The rule matches when the Condition you select matches, so any check you can express becomes a protection.
    • Condition Group: The rule matches when the Condition Group you select matches, which lets a combined set of conditions act as one protection.
    • Content Safety Hate: This detects hate speech directed at people or groups.
    • Content Safety Insult: This detects insulting or demeaning language.
    • Content Safety Racism: This detects racist content.
    • Content Safety Sexual: This detects sexual content.
    • Content Safety Violence: This detects violent content, including threats.
    • Content Safety Misconduct: This detects content that promotes illegal or harmful activity.
    • Denied Topic: This blocks the topics you list, comma separated. For example: “politics, medical advice”.
    • Pii Email: This detects email addresses.
    • Pii Phone: This detects phone numbers.
    • Pii Address: This detects physical addresses.
    • Pii Credit Card: This detects payment card numbers.
    • Pii Social Security Number: This detects social security numbers.
    • Pii Date Of Birth: This detects dates of birth.
    • Pii Username Password: This detects username and password combinations.
    • Pii Keys And Tokens: This detects API keys, access tokens, and similar secrets.
    • Pii Custom Sensitive Pattern: This detects the custom pattern you provide, so a sensitive format specific to your business is covered too.
    • Prompt Injection: This detects attempts to override your instructions through the prompt.
    • Custom Prompt: This screens content with a rule you write in plain language as a prompt, judged by the inferencer you choose.
  • Memory (Memory’s ID): This determines whether the Guard Rail keeps conversation context across requests. When a Memory is connected, relevant conversation history is stored and provided as context when the Guard Rail handles later requests.
  • Wiper (Wiper’s ID): This determines whether stored data associated with the Guard Rail is cleaned up according to a defined policy. A Wiper can remove data based on time or count limits.
  • Log Group (Log Group’s ID): This determines where the Guard Rail’s execution logs are recorded. When a Log Group, defined in Namirasoft Log, is connected, request activity and errors are recorded there for debugging and review.
  • Description (String): This is optional explanatory text about this Guard Rail. Use it to describe the protections it applies and where it is attached.
  • Created At (DateTime): This is the date and time when this Guard Rail was created. This value is automatically generated and cannot be modified.
  • Updated At (DateTime): This is the date and time when this Guard Rail configuration was last modified. This value is updated automatically whenever any field is changed.

Ready to Configure Your Model or Want to Explore the Console Guide?

Get Started Explore Console Guide